Token observation to account action

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides instructions and examples for tool-calling patterns involving authentication tokens. It ensures tokens are correctly extracted from tool outputs and passed to subsequent actions like querying a balance or modifying a password.- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the flow of data from tool observations into tool parameters. This pattern represents a surface for indirect prompt injection if an external system provides a malicious token, but it is a standard functional requirement for token propagation.
  • Ingestion points: The skill processes data from the output () of the GetUserToken tool.
  • Boundary markers: Instructions explicitly reference the delimiter for extracting tokens.
  • Capability inventory: The skill interacts with account-specific tools such as QueryBalance, DeleteAccount, ModifyPassword, and AccountInfo.
  • Sanitization: Parameters are copied directly as required by the tool schemas.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 05:13 PM
Security Audit — agent-trust-hub — Token observation to account action